The purpose of the current paper is to explain how one can forecast the effect of an elected tax on saturated fat on the demand for butter. The tax is to take affect from the first of January 2010 in Denmark. The tax is supposed to affect the consumption of saturated fat and especially high consuming households are of interest. Quantile regression is thus better suited than mean regression. Interest centre on at risk groups with larger consumption, but we are also interested in a simple measure that measure the total effect of the tax change, i.e. the unconditional quantile. The former can easily be obtained from the quantile regression while it is proposed to use simulations in the latter case. In mean regression a close form formula for calculating the unconditional mean from the conditional mean exist; unfortunately this is not the case for quantile regression. Hence, simulations are needed. The principle in the proposed method is the same as the methodology used in a recent published paper for comparing labour income distributions. A refinement of this methodology is suggested.